在多元種族的國家裡，族群群聚是很普遍的現象。一個簡單的解釋是人先天偏愛同類者，然而，這個說法並非為造成群聚的必要條件。本文說明即便多數族群僅根據少數族群的犯罪率高低決定歧視與否，而非根據外表差異，少數族群仍舊傾向聚集。就算各族群的犯罪率皆相同，在資訊不完全下，多數族群仍可能因誤判而歧視少數族群，而造成這種誤判的原因乃是因為樣本統計誤差所致。因此，少數族群適度群聚能減低被歧視的機率，從而形成特定的人口分佈型態。最後，文中也探討其他相關參數如何影響聚集的程度，並用模擬的方式驗證。

Racial clustering is common in a multiracial society. One possible explanation is that people prefer their own kind. This paper constructs a model in which the majority group discriminates against the minority group purely based on the latter's crime rate instead of other outer characteristics such as race or skin. We find that even though the crime rates of all groups are equal ex ante, the minority group still tends to cluster. This is so because the majority group could still discriminate against the minority group due to misjudgments under incomplete information. Such misjudgments come from the high variation of crime rates in small samples. Therefore, the minority group can reduce the probability of being discriminated against by clustering to reduce the variation of its crime rate. We also illustrates the effect other relevant parameters have on the minority group’s clustering behavior and test the results with simulations.